| With the construction of infrastructure and the need of urban development,many high fill projects have appeared in the western area of loess.In the construction process,due to the difference of the filling material,compaction technology and management process,the compaction quality of the filling soil will appear varying degrees of variability in the space range.However,the properties of fill soil are closely related to the compaction quality,and its variability has an important influence on the safety of fill slope and corresponding infrastructure construction.Existing studies usually use random field theory to analyze the influence of such variability on slope stability and reliability,but there are few studies on slope deformation prediction.The long-term deformation of fill slope is caused by creep of fill soil.Therefore,on the basis of analyzing the existing deformation prediction methods,this paper will study the creep characteristics of loess through one-dimensional consolidation creep test,establish the corresponding creep model,use the random field theory to characterize the creep variability,and use the finite element software to predict the deformation of high sticking slope.The main research contents and achievements are as follows:(1)This paper briefly introduced two kinds of common slope deformation prediction methods: theoretical type and statistical type.Through the analysis of examples,the advantages and disadvantages of hyperbolic method,ARIMA model and BP neural network model are compared.Secondly,the theoretical method combining creep theory and finite element numerical simulation is the most suitable method to consider soil variability.(2)The effects of dry density and vertical pressure on the consolidation creep characteristics of loess were summarized by one-dimensional consolidation creep test,and the consolidation creep variability of compacted loess under reproducibility test was analyzed.A semi-empirical and semi-theoretical creep model with simple parameters and high applicability is derived,and the parameter sensitivity analysis is carried out.Secondly,the UMAT subroutine of the creep model is written using Fortran language,and the ABAQUS numerical model is established to verify the correctness of the subroutine.(3)The random field theory is briefly introduced.The correlation of creep parameters is analyzed and the probability distribution model of Gaussian distribution of creep parameters and its parameters are established.A random sample matrix of creep parameters is derived,and a Gaussian random field of creep parameters is established.By combining the creep parameter random field model with the finite element method,and through the secondary development of ABAQUS based on Python,a procedurally operated method for predicting the deformation of high sticking slope considering the creep randomness is proposed.(4)On the basis of analyzing the general situation of the project,the finite element model of the high pasted slope is established,and the deformation characteristics of the high pasted slope are analyzed by using the loess creep model.The measured deformation data are used to verify the creep characteristics of long-term deformation of high sticking slope,and the proposed MC-SFEM is used to predict the deformation.The measured deformation is basically within the predicted deformation range,and the predicted deformation mean is close to the hyperbolic method and BP neural network model,which verifies the rationality and reliability of the proposed method.On the basis of predicting the deformation,the probability distribution characteristics and variability of the deformation after high sticking slope construction are analyzed。... |
PDF Full Text Request